Generalization of cardiomyopathy classification models using image feature descriptors from Magnetic Resonance Imaging

Abstract

Combining Image Processing and Artificial Intelligence techniques has proven to be a valuable resource for solving problems in the healthcare field, especially regarding Computer-Aided Diagnosis (CAD) based on medical images. In this context, one of the issues currently explored in research conducted at the Laboratory of Health Informatics Applications (LApIS) is the support for diagnosing cardiomyopathies, which are cardiac abnormalities that can be detected by investigating left ventricular characteristics in Cardiac Magnetic Resonance Imaging (CMR). The proposed Scientific Initiation research has the general objective of evaluating the generalization capacity of classification models based on Supervised Machine Learning (SML), applied in the context of CAD for cardiomyopathies, built and tested in previous work. These models were constructed using a set of individual image descriptors and CMR image sequences developed at LApIS in previous research. The methodological phases planned for the research include: conceptual studies and literature review; experiment modeling; implementations; experiment execution; and analysis and summarization of the results through the writing of materials for scientific dissemination. It is expected that the research results will contribute to defining high-performance classification models that can be reused to build a complete software infrastructure to aid in the diagnosis of cardiomyopathies.